I don’t think the on-demand model is necessarily profitable - that depends on good management, appropriate pricing and so on. It’s my (obvious) guess, however, that Kitely has been designed and managed to be profitable and thereby ensure its continuation.
What I find more noteworthy about their service structure is that it is VERY economical for users. Such win-win situations in business are role model material. As you note, however, and like any other service in any other industry, there must be a good fit between the service offering and the customer’s needs in order to trigger a transaction.
Counting regions is a metric that will probably (hopefully) evolve over time. What I like about Kitely is that they have a public directory which offers a more meaningful aspect of their worlds, namely descriptions that can help the user decide, “Is this world going to be interesting to me?”